Executable Threshold–Temporal Rules over Clinical Time Series: scoping review protocol v1.0

Abstract

Protocol for a scoping review of executable threshold–temporal rules over clinical time series: rules that contain both a numerical threshold and a temporal operator, applied to sequences of clinical or physiological observations. The review charts four questions: (RQ1) which formalisms are used and at which stage of the rule lifecycle; (RQ2) where the threshold and where the temporal parameter come from, separately from each other and separately from the origin of the rule structure; (RQ3) what kind of evidence is offered and at what level — formula, parameter, or prediction; (RQ4) how validation related to the development data, and whether the rule was modified in the process. Conducted according to the JBI Manual for Evidence Synthesis (ch. 10) and reported according to PRISMA-ScR. The protocol is not registered in a review registry; PROSPERO does not accept scoping reviews. This version was frozen after completing the novelty check, the construction of two control sets, the search pilot and the coding-form pilot, and before any formal screening. Section 13 lists all nineteen changes introduced during those pilots, each with its reason and the measurement that justified it — including the removal of one search term that accounted for 99.1% of a stream's yield, and the removal of a threshold-and-duration filter that reduced control-set coverage to zero because that vocabulary occurs in Methods sections rather than in abstracts. Search strategy coverage against a 15-paper control set: 92.3% overall, 90.9% for the phenotyping subset.